AI Lab Mirendil Raises $1B Series A: Release Names No Valuation, Cites Only Company Tests
Reported by Tech Times1 source ↓
Mirendil announced a $1 billion Series A on Nov. 12, led by Kleiner Perkins and Clearlake Capital. The announcement did not state a valuation.
Key points
- The round came 141 days after Mirendil raised $200 million at a $1 billion valuation
- Mirendil says its smallest model cost less than $10 million to train
- The reported model comparison was run by Mirendil, with no independent evaluation reported
Details
Tech Times reports that the $1 billion Series A announcement gave no valuation. Bloomberg had reported in September that Mirendil was in talks to raise up to that amount at a $5 billion valuation, but Tech Times says no source confirms that those were the final terms.
The less-than-$10-million training figure covers Mirendil-Base, the smallest model in the company’s family. Mirendil says larger models were still training when it published the figure on Oct. 21, and Tech Times notes that the company has not disclosed token counts, active-parameter counts or GPU hours.
Mirendil reported a comparison on seven code repositories published after its models were trained. Its Base model averaged 0.222 bits per byte, behind Kimi-K3 at 0.193, where lower is better; Tech Times says the company ran the test and no independent party had reproduced it. The article also notes that this prediction metric does not establish performance on coding tasks, reasoning or tool use.
Written from coverage as of
Images
- Tech Times